Originally Posted by Drek
****in' hell man. You really think some scrub from the National Review is going to game Nate Silver?
People got a lot of learning to do about statistics. Fact is, Silver's model is likely being swayed by fickle state polls that are currently oversampling the right under pretenses that the right is more "energized". He's still winding up with a strong chance of Obama winning.
The right this year sounds A LOT like the left did in 2004. "Nah uh! our guy closed the gap big time after that first debate. You wait and see! Them polls are wrong, its' way closer, I can feel it!"
Yeah, sure. Talk to you on election night at about midnight when Ohio is called blue. That coming after Wisconsin, Iowa, Virginia, and likely New Hampshire are called blue before it.
Obama is ahead in state polls and has by far the better ground game. The GOP's only hope is massive voter disenfranchisement.
No I just think Silver is a human, prone to make mistakes that other humans make. I find the question of why he weights certain older polls more than newer others... interesting. Why do all of his weights go in one direction? These are questions I'd like Silver to answer, among others.
Obama may still win, but the differences from 2004 are numerous. Just a couple - Bush was above 50% in the Gallup poll right at this same time, Obama is not. The economy was growing about twice as fast in 2004 than now.
I prefer RCP poll average
(link to national avg, drawback is that there are less state polls available, then again, the state polls really only matter if the election is within about 1.5%, and yes, it might be that close), best thing to overcome poll samples is to get larger responses. It's an issue I don't know how the polling companies overcome given that only 9% of people they call even agree to participate. RCP has Romney up 0.4% versus 2004
when Bush had a 2.4% advantage.
So no, I don't take Nate Silver and his calculators better than any other prognosticator.